Originally posted by normhead I took my numbers form Imagine Resources. Photozone probably hasn't tested a Pentax product in over two years and test a paltry number of lenses last time they did test which was on a K-5. Imagine resources at least tries to test all cameras with the same lens to try and come up with some kind of testing consistency. Your method really doesn't tell us anything but who makes the best lenses for their camera. You may not like my numbers, but that in no way means yours are better. You might want to do a bit more research on this. I used to rely heavily on photozone, and still do, for lens characteristics. But they've never tested a lens on a K-5ii, a K-5IIs or a K-3. Their info is getting to be a little dated. Citing a D7000, now that's really dated.
From Imagine Resource
Sony NEX-7 Review - Exposure Pentax K-3 Review - Exposure
I tend to use Imaging Resources for this kind of thing because they use the same lenses, where possible the Sigma 70 macro, so you're comparing sensors. not lenses.
K-5ii
Pentax K-5 II Review - Exposure
2300 lw/ph horizontal compared to 2850 is a 23% increase.
I like Klaus' work over at Photozone but for comparative sensor's it's not an appropriate format. At least the folks at IR have made an effort to create a level playing field for sensor comparisons. You can debate how effective they've been, but at least they tried. Klaus makes no such effort.
Keeping the same lens from sensor to sensor work only if the lens outresolve by a huge marging all sensor so the lens quality do not play any role in the comparisons.
If some lens manage to get 3675 on a Nex, then the sensor can go up to that. One example is enough if the example is accurate to know the sensor can at least go up to that level.
Now funilly we go from 2300 to 2850 instead of 2200 to 2500. that still amost double the difference you compute in your computations.
I guess the number difference is how image resource and photozone put the limit. Photozone get higher number but they are consistent, like apparently imaging resource. What would be wrong then is to give lw/ph without citing then the source or compare between different sources.
As for K5 vs D7000 vs K5-II they all use the same sensor so there no real point to say this old blabla. This is same technology, should get similar results.
Still I think this is not the right way to approach it. I still think that a K5 if you keep all the pixels is already pretty good up to really large sized print. At 40" the 0.4mm you compute indicate something not really satisfactory if you look near but that not typically you do out of a 40" print. Even through, honestly if I were to do 40" and more prints all the time I would have think serioulsy of getting a D800E/D810/D800 and if made a living out of that, of a 645Z. That because I suspect I might do a 30" or 40" print only from time to time if any that I don't care much what the K5 does in thoses settings. Knowing that it will more than good enough if you don't look from near distance is more than good with me.
Now the question is how often you crop, how much and what size you'd want to get your prints. If one is heavily cropping (like for wildlife), he might get thoses 0.4mm results not at 40" print that are pretty uncommon, but 20" that is much more common. This is the size of an Art book double page, or of a 24" monitor screen. There 0.4mm would be more visible and a little more disapointing.
So it is just a question of what you expect, what you do, not of absolutes.